Hi, Welcome
.?.:
Faster 1.0
Steady

YASH GUPTA

Applied Machine Learning & Backend Engineer

Yash Gupta

My perspective on building intelligence and pipelines in production.

Predict. Optimize. Deploy.

Scroll

Building ML models is a solved problem.
Integrating them into low-latency production pipelines is where real product value is unlocked.

System design for high-throughput concurrency.

Low Latency.
High Throughput.

RAG, embeddings, and vector databases.
Infrastructure tools are only as powerful as the indexing and retrieval strategy behind them.

Model selection, parameters, and fine-tuning datasets.

Do we need code..?
or solutions?

Airflow pipelines running 500+ daily runs.
Production workflows must be self-healing, highly observable, and failure-resilient.

Continuous validation testing and monitoring loops.

Reliability comes from consistency.

Based in Amsterdam, NL.
Developing computer vision and telemetry microservices impacting millions of sessions.
Ready to scale your models? Let's connect..

Bridging intelligence and business metrics.

You be the perception,
I'll be the tool.

yashgpt-inference-pipeline.log

Featured Projects

📺

Video Skip Intro/Outro Intelligence

An end-to-end computer vision and audio matching microservice to automatically identify skip markers in streaming content, scaling across millions of active playback sessions.

ResNet-3D Audio Spectrogram FastAPI Kubernetes
View Repository
📈

Low-Latency Inference System

Low-latency inference server serving models with sub-15ms P99 responses. Integrated Docker, Kubernetes scaling pods, and Grafana validation dashboards for extreme resilience.

PyTorch Docker Prometheus Grafana
View Repository

Network Resilience Simulator

A backend framework designed to emulate chaotic network conditions and high-concurrency request loads to test the stress limits of critical microservices.

Python Load Testing Async I/O Kafka
View Repository
🧠

RAG Knowledge Pipeline

A retrieval-augmented generation system with vector search, embedding pipelines, and LLM orchestration for intelligent document Q&A at enterprise scale.

LangChain Embeddings Qdrant GPT-4
View Repository

GitHub Repositories

portfolio

Personal portfolio website — built with vanilla HTML, CSS & JS, featuring interactive ML pipeline simulations and terminal.

JavaScript
View on GitHub
sppu-comp-be

University coursework and backend engineering projects from Computer Science curriculum.

Python
View on GitHub
sudoku

Sudoku solver using backtracking algorithm — a clean implementation of constraint satisfaction.

Python
View on GitHub

Professional Experience

Technology Lead / Applied ML Engineer

Infosys 2021 – Present
  • Scaled ML-powered product features across 7 European markets, impacting high-volume consumer applications.
  • Designed and automated Airflow pipelines handling 300–500+ DAG runs/day across batch and real-time inference.
  • Productionized ML models into low-latency inference APIs for seamless product integration.
  • Developed computer vision/video intelligence features (skip intro/outro) across millions of playback sessions.
  • Improved system reliability and SLA tracking via SLIs/SLOs, monitoring dashboards (Prometheus/Grafana), and validation frameworks.
  • Built resilient infrastructure using Docker and Kubernetes, ensuring fault tolerance and high availability.

Senior Systems Engineer

Infosys 2018 – 2021
  • Designed and developed high-performance backend systems and APIs for data-intensive applications.
  • Built a robust traffic simulation and network emulation framework for resilience testing under extreme network load.
  • Improved legacy system performance and scalability through modernization, caching, and code refactoring efforts.

Systems Engineer

Infosys 2016 – 2018
  • Developed automated data processing and engineering pipelines to support downstream analytics and ML workflows.
  • Optimized SQL database query performances and built schedulers for ETL batch processing.

Education

🎓

B.Tech, Computer Science

SRM Institute of Science and Technology

About Me

With over 8 years of experience in engineering, I specialize in the convergence of machine learning models and high-performance backend systems. I bridge the gap between data science and product engineering, translating complex research models into performant, production-ready microservices.

Currently based in Amsterdam, Netherlands, my focus lies in refining real-time inference systems, designing large-scale Airflow pipelines, and implementing retrieval architectures (RAG) using modern LLMs and vector search.

I build software with the assumption of scale, resilience, and change.

0+ Years Experience
0 EU Markets
0+ Daily DAG Runs

Skills Grid

Applied Machine Learning

ML Inference Low-Latency APIs Feature Pipelines Model Validation Video Intelligence

Generative AI & Search

RAG Architecture Vector DBs (Qdrant/Pinecone) Embeddings LLM Orchestration

Backend & Distributed Systems

Python (FastAPI/Flask) Distributed Systems REST & gRPC APIs Kafka Airflow SQL

Cloud & Observability

Docker Kubernetes (K8s) AWS & GCP Prometheus Grafana ELK Stack

Get In Touch

I am currently open to new opportunities, technical discussions, or collaboration projects. Reach out via email:

yashgpt2894@gmail.com

Interactive Terminal Console

yash-guest@portfolio:~
Welcome. Type help to begin exploring. Try running inference or skills!
guest@yash-ml:~$